motivons

Motivons are hypothetical discrete units of motivational energy used in theories of decision making, psychology, and artificial agents to quantify an agent’s propensity to engage in goal-directed behavior. The concept treats motivation as a modular resource that can be accumulated, allocated among tasks, and depleted by effort or time.

Origins and concept

Motivons extend ideas from drive theory and intrinsic motivation by modeling motivation as separable from immediate reward signals. In many formulations, motivons are scalar quantities that reflect an agent’s current readiness to act toward a goal, influenced by expected value, effort costs, and contextual factors. They are designed to be dynamic and context-sensitive, changing as the environment and internal state evolve.

Dynamics and mechanisms

Motivons interact with action selection by biasing choices toward higher-valued goals. In simple models, the motivational level M changes over time according to factors such as reward expectations, anticipated effort, and decay. A common abstraction is a bounded, time-varying value that increases with favorable predictions and decreases with costs, fatigue, or competing goals, thereby modulating vigor and persistence of behavior.

Measurement and applications

Motivons are inferred from behavioral data, such as choice probabilities, response times, and action vigor, or estimated within computational models by fitting to observed behavior. They are used in psychology to explain shifts in motivation and in artificial intelligence to separate intrinsic motivation from external rewards. In game design and education, motivon dynamics can help regulate pacing and challenge.

Criticism and alternatives

As a theoretical construct, motivons risk overparameterization or circular definitions. Critics urge careful empirical validation and emphasize that motivation is multifaceted, potentially better captured by multiple interacting processes rather than a single unit.

See also

Motivation; Drive theory; Intrinsic motivation; Reinforcement learning; Computational neuroscience.